Objective: A previous meta-analysis concluded that TNF-alpha 238A/G and TNF-alpha 308A/G polymorphisms were not associated with the risk of juvenile idiopathic arthritis (JIA) in the overall population or Caucasian subjects. With the publication of a fair number of studies on the association between TNF-alpha polymorphisms and JIA in recent years, we conducted this updated meta-analysis to make a more accurate evaluation of such relationship. Methods: We adopted PubMed, EMBASE, ISI Web of Science and CNKI to identify observational studies that addressed the association between TNF-a polymorphisms and risk for JIA. The allelic effect of variant A for the risk of JIA was expressed as odds ratio (OR) along with the associated 95% confidence interval (95% CI). Meta-analyses were performed by pooling ORs and 95% CI from included studies using RevMan 5.3 software. The stratified-analysis based on ethnicity was performed to confirm the ethnicity-dependent effect on the relationship. Results: A total of 15 case-control studies including 2845 patients in JIA groups and 4771 patients in control groups were included in our study. The findings indicated a statistically significant association between the A allele of the TNF-alpha 238A/G polymorphism and the decreased JIA risk in Caucasians (P=. 0002). The study in Iranian showed similar results (P=. 0002) whereas the studies in other ethnicities failed to replicate this finding: Han (P=. 29), Mexican (P=. 64) and Turkish population (P=. 32). TNF-alpha 308A/G was not statistically associated with JIA in overall subjects or Caucasians. Conclusion: Our study confirmed the protective role of the A allele in TNF-alpha 238A/G but not TNF-alpha 308A/G against the occurrence of JIA in the Caucasian population. To exactly validate the correlation between TNF-alpha polymorphisms and JIA in other ethnic backgrounds, additional studies are required.